OBJECTIVE Patients with latent autoimmune diabetes in adults (LADA) express autoantibodies against the 65-kDa isoform of GAD antibody (GADA). Intervention with recombinant human GAD65 formulated with aluminium hydroxide (GAD-alum) given twice subcutaneously to LADA patients at intervals of 4 weeks was safe and did not compromise β-cell function in a Phase II clinical trial. GADA affinity has been shown to predict progression to type 1 diabetes. Here, we asked whether GADA affinity was affected by the GAD65 antigen-specific vaccination and/or associated with β-cell function in participants of this trial. RESEARCH DESIGN AND METHODS GADA affinity was measured in sera of 46 LADA patients obtained prior to the 1st week and 20 weeks after the second injection with GAD-alum or placebo using competitive binding experiments with [125I]-labeled and unlabeled human GAD65. RESULTS At baseline, GADA affinities ranged from 1.9 × 107 to 5.0 × 1012 L/mol (median 2.8 × 1010 L/mol) and were correlated with GADA titers (r = 0.47; P = 0.0009), fasting (r = −0.37; P = 0.01) and stimulated (r = −0.40; P = 0.006) C-peptide concentrations, and HbA1c (r = 0.39; P = 0.007). No significant changes in affinity were observed from baseline to week 24. Patients with GADA affinities in the lower first quartile (<4 × 109 L/mol) had better preserved fasting C-peptide concentrations at baseline than those with higher affinities (mean 1.02 vs. 0.66 nmol/L; P = 0.004) and retained higher concentrations over 30 months of follow-up (mean 1.26 vs. 0.62 nmol/L; P = 0.01). CONCLUSIONS Intervention with GAD-alum in LADA patients had no effect on GADA affinity. Our data suggest that patients with low GADA affinity have a prolonged preservation of residual β-cell function.